Dear Apart Community,

Welcome to our newsletter - Apart News!

At Apart Research there is so much brilliant research, great events, and countless community updates to share.

This week's scorching edition of Apart News begins with our Agent Security Hackathon kicking off this weekend, our new and exciting Jam Site locations announcement, a whole host more of paper submissions shipped by Apart Lab, and updates on our ongoing Apart Balearic Retreat.

Ensuring Agent Security

*Our Agent Security Hackathon has kicked off tonight!*

If you didn't get to be part of this Apart Sprint (first of all we're sorry to hear this because they're great!) make sure you never miss one again by signing up here.

While most AI safety research focuses on LLMs, the AI systems being deployed in the real world are far more complex. Which is why we are running the Agent Security Hackathon: because the increasing complexity of agents necessitates a reevaluation of existing safety frameworks and the development of novel approaches to security.

Our CEO, Esben, writes in his Agent safety research article that AI agents are 'computing systems with a brain designed to orchestrate actions from perception.' So it is crucial we ensure they are developed safely.

Those participating in the Agent Security Hackathon and other Sprints will: gain experience in cutting-edge AI safety research; develop valuable skills in AI development, security analysis, and problem-solving; network with leading experts; contribute to solving pressing challenges in AI development; enhance their resume/CV; kickstart a career in AI safety.

Team Offsite: Apart Balearic Retreat

Our team offsite brought together everyone at the same time (including our new colleagues) for the first time this week in sunny Spain.

Although the Wi-Fi could work better, we have been talking about everything from research directions to what the word in each of our home countries is for a 'living room.'

Last week, Apart’s new co-authored paper, Interpreting Learned Feedback Patterns in Large Language Models, has been accepted to the prestigious NeurIPS.

Whilst on the offsite this week, we also finished and submitted a total of four papers to ICLR. And we had six unique submissions to NeurIPS and eight in total; two of which were submitted to two different workshops

New Jam Sites

Our Jam Sites partners are the physical locations where our Apart Sprint participants can spend time on their pilot experiments in AI Safety. So give a huge welcome to our new, exciting Jam Sites at CEELAR Hotel and Effective Altruism Singapore.

AI Safety Asia

Esben was invited to speak at AI Safety Asia's inaugural roundtable on South-East Asian technical AI safety policy.
​
He was joined by Robert Trager, Co-Director of the Oxford Martin AI Governance Initiative, International Governance Lead at the Centre for the Governance of AI (GovAI), and Senior Research Fellow at the Blavatnik School of Government, University of Oxford; the former Indonesian Minister, Gita Wirjawan also joined; alongside the President of KORIKA, Hammam Riza.
​
Thanks to Edward Tsoi and Lyantoniette Chua, AISA's Co-Founders. AISA is a 'non-profit organization dedicated to AI Safety and Governance talent development and upskilling across Asia.'

AISA intend to 'reduce catastrophic AI risks by identifying and closing gaps on policy levers and transnational coordination for safe and governed AI, starting in Southeast Asia.'

Talk at ML4Good

Esben also gave a talk and Q&A at ML4Good. ML4Good hosts intensive in-person bootcamps to up-skill participants in AI Safety.

Opportunities

• Future Hackathons: remember to sign up here for future sprints.

Have a great week and let’s keep working towards safe AI.

‘We are an AI safety lab - our mission is to ensure AI systems are safe and beneficial.’